lkml.org 
[lkml]   [2012]   [May]   [4]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: [PATCH] sched: Enable arch-specific asym packing option in sched domain
Date
Diwakar Tundlam <dtundlam@nvidia.com> wrote:

> >> A repost doesn't make it right to add sibling (SMT) properties to a package/socket and will still upset Power7.
>
> Agreed. I only fixed the whitespace and commitId.
>
> arch_sd_sibling_asym_packing() is already present under ifdef CONFIG_SMT.
> I didn't touch that. I only added it to SD_CPU_INIT for all cpu's.
> I assumed Power7 shouldn't use SD_CPU_INIT.

You are using sibling at the cpu level. POWER7 sets this at the sibling
level and we don't really want it at the cpu level.

> I don't understand Power7 arch to comment on impact.
> Michael Neuling should review this carefully and advise.

Ask Peter said, it's broken for POWER7. We don't want to set this at
the cpu level.

> Maybe I should define a separate weak symbol, say
> arch_sd_bias_to_lower_num_cpu()? Then Power7 can define
> arch_sd_sibling_asym_packing() to be '1' and it will not break all-cpu
> init.

This sounds better but I'd follow the old name call it
arch_sd_cpu_asym_packing()

Mikey


\
 
 \ /
  Last update: 2012-05-05 01:21    [W:0.169 / U:1.300 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site